News & Analysis
Analyst rails against the fab-lite
Peter Clarke
1/28/2010 11:21 AM EST
LONDON — The fabless chip vendors and pursuers of fab-lite manufacturing strategies are about to get a shock, according to Malcolm Penn, principal analyst with Future Horizons (Sevenoaks, England). Some companies could under-perform, others could be driven out of business in a market that should be booming for them — and all because they have lost control of manufacturing.
According to Penn the fab-lite business model is: structurally deceitful, operationally faulty, and financially flawed. "Fab-lite is yet another bean-counting financial analyst deception, just like the disgraced private-equity 'debt is good' business model. Going IDM to fabless does not solve the underlying problems. It is simply dicing with death," Penn told attendees at a one day seminar on the state of the global semiconductor market.
Penn pointed out that outsourcing chip companies will, at the very least see wafer prices increasing. At worst they may miss market windows for lack of chips. "In many ways allocation is a bigger problem than prices increasing," said Penn.
The background to Penn's commentary is that, with a lack of investment in manufacturing capacity over the last three years, ASPs are set to rise at the same time as the general economy is recovering. It is Penn's contention that the worldwide chip market is set for two years of more than 20 percent growth and could even hit more than 30 percent growth in a single year (see Future Horizons sees 30% chip market boom).
"The only people building fabs are Intel, Samsung and a few foundries," said Penn, highlighting one reason why he believes the semiconductor industry is about to go into short supply.
The fact is that a whole swathe of the semiconductor industry, including such giants as Texas Instruments and STMicroelectronics are opting to go fab-lite. For many companies that means continuing to manufacture older and analog products in legacy fabs but avoiding the massive cost of investing in a new wafer fab. For leading-edge digital CMOS these companies must therefore outsource their needs to one of relatively few foundry sources.
Comments
Peter Clarke
1/29/2010 6:56 AM EST
Malcolm Penn has always been in the fabbed-is-good camp and distrustful of the fab-lite business model. But can thousands of finance people be wrong?
Is fabless and fab-lite only a good strategy in times of plentiful wafer supply? Or will past supply agreements and contract price-fixes keep the fab-lite and the fabless accessing their silicon at reasonable prices?
If you have experience of the situation or deep conviction let us know what you think.
Sign in to Reply
stinky08
1/29/2010 7:22 AM EST
There is always a economical scale to be paid attention on. If you have a big fab and your business is barely covering the cost then you better let it be. Fab-lite is a better approach since you can leverage the foundry and do not get chocked by foundry when things get rough! Fab-lite gives you also the possibility to preserve your knowhow and uniqueness!
Sign in to Reply
ScottenJ
1/29/2010 11:37 AM EST
ASP is well correlated with Utilization on a year-to-year basis. Due to the under investment Malcolm Penn has discussed utilization will likely be very high next year and I agree that ASPs will likely rise. My modeling for 2010 suggests greater than 20% growth. I havenât looked at 2011 yet but if underinvestment continues it will likely be another strong year.
In terms of Fab versus Fab-Lite, I also think the economics are misunderstood by most. I think fabless makes sense for certain business segments but not for all. If you can build and keep a 300mm fab full of sufficient scale to be economical, then instead of paying a foundry a margin you can keep the margin for yourself. But you have to keep it full and have enough âscaleâ. You also have to deal with the relative advantages and disadvantages of different locations.
Sign in to Reply
ScottenJ
1/29/2010 11:39 AM EST
Just to clarify my last comment, greater than 20% growth in revenue, not ASP driven by growth in units and ASP.
Sign in to Reply
ssee
1/29/2010 12:50 PM EST
To answer Peter Clarke's specific question about "can thousands of finance people be wrong?"
Yes, thousands of people can be wrong. Look at the financial sector meltdown. All the quant and math wiz kids got a unsustainable model and for years it seemed like it was working fine.
Both fab and fab-lite have room to co-exist. It depends on what product space that the company is trying to address with that model.
Sign in to Reply
djafer
2/1/2010 12:17 AM EST
The main assumption of the author is that allocation will cause a durable wafer price increase. Allocation will be gone in a quarter or two once market adjustments would have occured. However, the competition in the field of semiconductor foundries is still very tough, and the foundry offering has been steadily increasing during the last years, having more and more players around. Therefore, there is no real reason to see the fablite players suffering longterm from this allocation. In addition, major fablite players like STM or TI, will certainly get priority at the foundry side, whereas small design houses will be pushed at the buttom of the list.
Sign in to Reply